Check multiple ports under the same name

[aelnosu]

Would it be possible to check multiple ports of the same site with an OR relationship.

So, for example it will check the status of port 22 and 80 of domain.com and if either of the port is not responding, it will return a site is down message.

For latency, it will report the average latency of all ports.

This is for monitoring my mail server which have port for SMTP, IMAP and POP, but I want them to be shown under the same service.

[M1XZG]

@aelnosu, I’ve been running Upptime since April 2023, and it works more like a collection of independent monitors than a service made up of several checks. Each entry under sites has its own port, status, history and response-time figures.

You can monitor all the mail ports like this:

sites:
  - name: Mail server - SMTP
    check: tcp-ping
    url: mail.example.com
    port: 25

  - name: Mail server - IMAP
    check: tcp-ping
    url: mail.example.com
    port: 993

  - name: Mail server - POP3
    check: tcp-ping
    url: mail.example.com
    port: 995

That will tell you whether each port is reachable, but Upptime will display them as three separate services. Make sure they have different names or explicit slugs, otherwise their generated history and graph paths will collide.

One small terminology point: I think the behaviour you want is technically an all-required relationship rather than OR. If the mail server should be marked down when any one of SMTP, IMAP or POP3 stops responding, then the combined service is healthy only when all three checks pass.

Upptime’s TCP checker already makes five connection attempts to an individual port and records the average time of the successful attempts. It does not currently combine results from different sites entries or calculate an average across several ports. The site configuration currently accepts one port, and the TCP checking code processes that port independently.

Separate entries may actually be more useful when something breaks because the status page immediately shows which mail protocol has failed. If you specifically need a single “Mail server” row, one incident and an average response time, I don’t think Upptime currently has a configuration-only way to do that. It would need either a composite-check feature or a small external health endpoint that tests the required ports and returns one result.

A native implementation could perhaps let a service contain several checks, choose whether all or any must succeed, and select how their response times are combined. That seems like a useful feature, particularly for services such as mail servers that depend on several related ports.

[aelnosu]

Thank you for the clearfication. On my current deployment, each port is listed as a separate service. It is working pretty well, but still definitely interested in a native implementation that can check all ports under the same site so there is less clutter.